Fixes https://github.com/flutter/flutter/issues/36091.
Previously, a sendRequest will be sent even if the peer is closed during
a driver test. That will cause a time out without any error information.
Such issue is unreproducible on my Mac Book Pro, but 100% reproducible
on our new Mac mini (2018).
The closing issue is tracked in https://github.com/flutter/flutter/issues/36268
Additional to this fix, we should also patch the jason_rpc_2 so the peer
will throw exception if sendRequest is attempted while the connection is
closed.
**Test**:
tiles_scroll_perf_iphonexs__timeline_summary failed without this patch.
It will pass after this patch.
I'm not sure how to add a unit test for this. Please let me know if you have
some ideas.
This patch will generate the following warning log on the new Mac Mini
```
flutter: Observatory listening on http://127.0.0.1:50192/cZPDF4sW7MM=/
Installing and launching... 10.2s
00:00 +0: scrolling performance test (setUpAll)
[info ] FlutterDriver: Connecting to Flutter application at http://127.0.0.1:1069/cZPDF4sW7MM=/
[trace] FlutterDriver: Isolate found with number: 3684677742843303
[trace] FlutterDriver: Isolate is paused at start.
[trace] FlutterDriver: Attempting to resume isolate
[trace] FlutterDriver: Waiting for service extension
[info ] FlutterDriver: Connected to Flutter application.
00:00 +0: scrolling performance test complex_layout_scroll_perf
[warning] FlutterDriver: Instance of '_WebSocketImpl' is closed with an unexpected code 1005
[warning] FlutterDriver: Peer connection is closed! Trying to restore the connection...
00:10 +1: scrolling performance test tiles_scroll_perf
[warning] FlutterDriver: Instance of '_WebSocketImpl' is closed with an unexpected code 1005
[warning] FlutterDriver: Peer connection is closed! Trying to restore the connection...
00:20 +2: scrolling performance test (tearDownAll)
00:20 +2: All tests passed!
Stopping application instance.
```
## Description
Print actionable warnings if `flutter drive` (where most of our performance benchmarks come from) is run in debug mode and it tries to gather benchmarks using `traceAction`.
## Related Issues
https://github.com/flutter/flutter/issues/30625
## Tests
I added the following tests:
* drive_perf_debug_warning devicelab test
* add trailing commas on list/map/parameters
* add trailing commas on Invocation with nb of arg>1
* add commas for widget containing widgets
* add trailing commas if instantiation contains trailing comma
* revert bad change
For `fuchsia_compat.dart` Instead of using `runSync`, use `run` to avoid
deadlock when attempting to access specific resources like the Hub in Fuchsia.
The specific example is that in Fuchsia, the `find` command is
attempting to explore `out` which hasn't yet been serviced, as `find` is
blocking on it, causing a deadlock.
These are essentially self-inflicted race conditions. Instead of timeouts we're going to try a more verbose logging mechanism that points out when things are taking a long time.
When setting the Fuchsia logging function, it should happen before any
initialization code, as init can still cause warning/error/info messages
to get printed to logs. Since the default stderr/stdout fd's aren't
correct, this can cause a program to crash for unclear reasons.
* Remove many timeouts.
These are essentially self-inflicted race conditions. Instead of timeouts we're going to try a more verbose logging mechanism that points out when things are taking a long time.
* Get the attach tests to pass.
* Apply review comments from Todd
* More review comment fixes
* Put back the extended timeouts here now that I know why we have them...
Adds a Fuchsia compatibility function that allows for connecting to a
specific Isolate by name when running Flutter Driver on a Fuchsia
device. This will check over multiple Dart VM's in an attempt to find
an Isolate by its name.
* Fix TODO syntax.
* Clarify messages for some timeouts, to aid debugging.
* Increase some other timeouts that were a needlessly short, to reduce sources of flakes.
* Remove some more timeouts that were mostly redundant, to remove complexity.
* Minor style cleanup.
* Remove some dangerous traps (specifically, hide the explicit start/end times in TimedEvent since they shouldn't matter).
fuchsia_tester.dart still assumes Dart 1. Previously, it ran tests directly
from source, flutter_platform.dart automatically runs a kernel compile when
operating in Dart 2 mode, but this assumes a functional Dart SDK is available
in the artifacts directly, and fuchsia_tester.dart mocks out the artifacts
directory with an empty temp dir.
Remaining work is:
1. Get the frontend server building as a dependency on Fuchsia.
2. Patch fuchsia_tester.dart to use a valid Dart SDK and frontend server.
This also reverts migration to Dart 2 typedef syntax.
This reverts commit 6c56bb2. (#18362)
This reverts commit 3daebd0. (#18316)
Now that Dart 1 is turned off, reapplying my change to turn on the prefer_generic_function_type_aliases analysis option, and fix all the typedefs to Dart 2 preferred syntax.
Also eliminated the unused analysis_options_repo.yaml file and turned on public_member_api_docs in analysys_options.yaml.
No logic changes, just changing the typedef syntax for all typedefs, and updating analysis options.
Our style guide says the k's are not necessary, and it seems like a good idea to make all the code be consistent on this.
Only naming changes to private vars: no logic changes.
In FlutterDriver.scrollUntilVisible, pause momentarily after we kick off
the waitFor() call, in order to give that call a chance to complete
before we start scrolling if the element is on-screen.
This change adds Dart VM event support (listening for when a VM starts/stops by using a periodic heartbeat).
This also adds support to connect to a specific `IsolateRef` through the flutter driver, so that when an application spawns, it can immediately be driven (as shown in included example code).
We're seeing occasional test timeouts trying to call `FlutterDriver.connect()`.
Unfortunately, when the test is timed out at the test runner level, you don't
get a meaningful test failure or stack trace of where the timeout occurrred.
Your test harness also doesn't get to clean up, which can include things like
saving the device logs to see what was going on in the device.
Thus, this change adds timeouts in the places where we've most
commonly observed hangs during tests.
* All lists produced by JSON parsing are List<dynamic>. If more
speficic type is required then they need to be explicitly cast, e.g.
using castFrom helper;
* Function of type (ByText) -> Finder is not a subtype of
(SerializableFinder) -> Finder because ByText is in the contravariant
position;
* In Dart 2 typed(any) should be used instead of any in mockito based
tests.
This allows the scheduler library to depend on the services library
and the painting library to depend on the scheduler library without
the services library having to depend on the scheduler library.
While I was at it I also cleaned up some of the binding logic: the
licenses logic can now be overridden (and the test library does so),
and the image cache can now be overridden as well.
This patch reorganises flutter_driver's lib/src directory into three
subdirectories:
* driver: sources transitively imported/exported by flutter_driver.dart,
which defines the API used by driver tests executed on the host machine.
* extension: sources transitively imported/exported by
driver_extension.dart, which defines the API used to instrument a
Flutter app running on the target device for use with a driver test.
* common: sources common to both driver and extension code.
This makes rolling Flutter into repositories that use Bazel (or
Bazel-like build systems) significantly more robust, due to the simpler
glob patterns involved.
1. We want to measure wall-clock duration for the benchmarks, as opposed
to thread duration (e.g., waiting on a mutex should accrue time) and
'dur' is the metric to use for that.
2. On Darwin-based systems (macOS and iOS) 'tdur' is the result of a
mach syscall lookup to thread_info. This call returns unreliable data
on iOS. Chromium, for example, disables thread time support entirely
for iOS.
I got tired of drive-by spelling fixes, so I figured I'd just take care of them all at once.
This only corrects errors in the dartdocs, not regular comments, and I skipped any sample code in the dartdocs. It doesn't touch any identifiers in the dartdocs either. No code changes, just comments.
* only tap on widgets reachable by hit testing
* use FractionalOffset
* added tests
* check finder finds correct widget
* undo unintentional changes
* address comments
* style fix
* add Directionality in test
* fix analysis warning
The flakiness of commands_test is generally visible via:
```
drive:stdout: Expected: 'log: paint'
drive:stdout: Actual: ''
drive:stdout: Which: is different. Both strings start the same, but the actual value is missing the following trailing characters: log: paint ...
```
By returning the name of the test we will be able to understand if
another test is running by reading the `Actual` value
- Wait for full Hot Reload
- Wait for full Restart
- Fallback if checkHealth throws METHOD_NOT_FOUND.
We try to wait for the service extensions to be registered and retry.
* Revert "Make plugins add their repos to projects in the consuming app (#11447)"
This reverts commit abe1e2520b.
* Revert "Support for custom build types on Android (#11354)"
This reverts commit 87eec719e2.
* Revert "add a profile() method (#11443)"
This reverts commit 561d17a876.
* Revert "Fix documentation based on dartdoc's warnings (#11428)"
This reverts commit 6655074b37.
* Revert "Improve some docs around WillPopScope. (#11429)"
This reverts commit 58a28a2965.
* Revert "temporarily disable broken driver test in integration_ui (#11440)"
This reverts commit 764515ec78.
* Revert "style fix"
This reverts commit 00bfc86630.
* Revert "tests for waitFor/waitForAbsent"
This reverts commit 31d2ee9c0c.
* Revert "Always evaluate the finder in `driver.waitFor()` and `driver.waitForAbsent()`"
This reverts commit 11d7c79b99.
Mainly, this adds documentation to members that were previously
lacking documentation.
It also adds a big block of documentation about improving performance
of widgets.
This also removes some references to package:collection and adds
global setEquals and listEquals methods in foundation that we can use.
(setEquals in particular should be much faster than the
package:collection equivalent, though both should be faster as they
avoid allocating new objects.) All remaining references now qualify
the import so we know what our remaining dependencies are.
Also lots of code reordering in Flutter driver to make the code
consistent and apply the style guide more thoroughly.
The main purpose of this PR is to make it so that when you set the
initial route and it's a hierarchical route (e.g. `/a/b/c`), it
implies multiple pushes, one for each step of the route (so in that
case, `/`, `/a`, `/a/b`, and `/a/b/c`, in that order). If any of those
routes don't exist, it falls back to '/'.
As part of doing that, I:
* Changed the default for MaterialApp.initialRoute to honor the
actual initial route.
* Added a MaterialApp.onUnknownRoute for handling bad routes.
* Added a feature to flutter_driver that allows the host test script
and the device test app to communicate.
* Added a test to make sure `flutter drive --route` works.
(Hopefully that will also prove `flutter run --route` works, though
this isn't testing the `flutter` tool's side of that. My main
concern is over whether the engine side works.)
* Fixed `flutter drive` to output the right target file name.
* Changed how the stocks app represents its data, so that we can
show a page for a stock before we know if it exists.
* Made it possible to show a stock page that doesn't exist. It shows
a progress indicator if we're loading the data, or else shows a
message saying it doesn't exist.
* Changed the pathing structure of routes in stocks to work more
sanely.
* Made search in the stocks app actually work (before it only worked
if we happened to accidentally trigger a rebuild). Added a test.
* Replaced some custom code in the stocks app with a BackButton.
* Added a "color" feature to BackButton to support the stocks use case.
* Spaced out the ErrorWidget text a bit more.
* Added `RouteSettings.copyWith`, which I ended up not using.
* Improved the error messages around routing.
While I was in some files I made a few formatting fixes, fixed some
code health issues, and also removed `flaky: true` from some devicelab
tests that have been stable for a while. Also added some documentation
here and there.
* Manually fix every use of Point.x and Point.y
Some of these were moved to dx/dy, but not all.
* Manually convert uses of the old gradient API
* Remove old reference to Point.
* Mechanical changes
I applied the following at the root of the Flutter repository:
git ls-files -z | xargs -0 sed -i 's/\bPoint[.]origin\b/Offset.zero/g'
git ls-files -z | xargs -0 sed -i 's/\bPoint[.]lerp\b/Offset.lerp/g'
git ls-files -z | xargs -0 sed -i 's/\bnew Point\b/new Offset/g'
git ls-files -z | xargs -0 sed -i 's/\bconst Point\b/const Offset/g'
git ls-files -z | xargs -0 sed -i 's/\bstatic Point /static Offset /g'
git ls-files -z | xargs -0 sed -i 's/\bfinal Point /final Offset /g'
git ls-files -z | xargs -0 sed -i 's/^\( *\)Point /\1Offset /g'
git ls-files -z | xargs -0 sed -i 's/ui[.]Point\b/ui.Offset/g'
git ls-files -z | xargs -0 sed -i 's/(Point\b/(Offset/g'
git ls-files -z | xargs -0 sed -i 's/\([[{,]\) Point\b/\1 Offset/g'
git ls-files -z | xargs -0 sed -i 's/@required Point\b/@required Offset/g'
git ls-files -z | xargs -0 sed -i 's/<Point>/<Offset>/g'
git ls-files -z | xargs -0 sed -i 's/[.]toOffset()//g'
git ls-files -z | xargs -0 sed -i 's/[.]toPoint()//g'
git ls-files -z | xargs -0 sed -i 's/\bshow Point, /show /g'
git ls-files -z | xargs -0 sed -i 's/\bshow Point;/show Offset;/g'
* Mechanical changes - dartdocs
I applied the following at the root of the Flutter repository:
git ls-files -z | xargs -0 sed -i 's/\ba \[Point\]/an [Offset]/g'
git ls-files -z | xargs -0 sed -i 's/\[Point\]/[Offset]/g'
* Further improvements and a test
* Fix minor errors from rebasing...
* Roll engine
Rename State.config to State.widget
Rename State.didUpdateConfig to State.didUpdateWidget
Renamed all State subclasses' local variables named config to something else
After this patch, there are three major text input widgets:
* EditableText. This widget is a low-level editing control that
interacts with the IME and displays a blinking cursor.
* TextField. This widget is a Material Design text field, with all the
bells and whistles. It is highly configurable and can be reduced down
to a fairly simple control by setting its `decoration` property to
null.
* TextFormField. This widget is a FormField that wraps a TextField.
This patch also replaces the InputValue data model for these widgets
with a Listenable TextEditingController, which is much more flexible.
Fixes#7031
For measuring the Dart thread, we care about thread duration (tdur) rather than
wall duration (dur) because we don't want to count the time when the Dart
thread is descheduled (e.g., in preference to the raster thread).
Prior to this change, these benchmarks were mostly measuring whether the OS
decided to finish the Dart thread's time slice or hand over the CPU to the
raster thread to complete the visual part of the frame. Now we actually measure
the work done on the Dart thread.
* Small Flutter strong mode cleanup fixes.
These are cases where strong mode down cast composite errors
generally indicated cases that would performance or correctness
issues if Flutter code was run in a strong mode VM.
* Fix Command API so that it is always in terms of Map<String,String>.
* Fix typedef
This remove a very brittle aspect of flutter drive, whereby it would
assume a known port instead of explicitly finding out what it was.
Fixes#7692 and hopefully fixes the devicelab tests.
